Data Modeling for Distributed SQL

on YouTube ↗29m

" summary

Explains how YBDB (YugaByte) uses the primary key to determine sharding: choose hash sharding for write distribution and range sharding for range scans, since hash defaults to one tablet per node. Covers sequence cache tuning (default 100 per connection, can increase to 1,000 per tserver to reduce hotspots), tablet sizing costs (~1,000 tablets ≈ 0.4 vCPU) and avoiding unnecessary splits via colocation and split_in, and recommends using the Voyager migration tool with small test clusters, schema analysis, and tuning before full imports.

AI-written summary · about

" chapters (13)

1. 0:00Welcome
2. 1:08Distributed vs traditional SQL
3. 2:57Data sharding with primary keys
4. 4:44Hash vs range sharding
5. 6:543 sharding design points
6. 7:16Primary key design methods
7. 9:27Secondary indexes in action
8. 14:20Table sizing and colocation
9. 17:15Best practices for models
10. 17:35Voyager migration tool
11. 21:07UUID anti-patterns vs sequences
12. 24:03Denormalizing data models
13. 28:03Summary and Q&A

" related talks

• 27mData Modeling: Master This Critical Step to Adopting a Distributed DatabaseJohn Schulz · Distributed SQL Summit 2022
• 30mPerformance: Tips & Tricks to Optimize the Speed of Your Mission-Critical DatabaseAlan Caldera, Bryan Whitmore · Distributed SQL Summit 2022
• 02h 27mDistributed SQL Tuning and Optimization in YugabyteDBFranck Pachot · Distributed SQL Summit 2022
• 01h 27mYugabyteDB YSQL DevelopmentSeth Luersen · Distributed SQL Summit 2022
• 27mIntroduction to YugabyteDB: Design and ArchitectureKarthik Ranganathan · Distributed SQL Summit 2020
• 29mEnterprise Scale: How a Distributed Architecture Can Deliver Scale and ResilienceAmit Chauhan, Rahul Batta · Distributed SQL Summit 2022
• 27mIntroduction to YugabyteDB – Design and ArchitectureKarthik Ranganathan · Distributed SQL Summit Asia 2021
• 01h 00mParallel Asynchronous Replication b/t YDB Database InstancesAndrey Fomichev, Ilnaz Nizametdinov · Hydra 2022